Journal of Membrane Science 392–393 (2012) 88–94

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science

journal homepage: www.elsevier.com/locate/memsci

Transport of liquid water through Nafion membranes

Qiongjuan Duan a , Huaping Wang a , Jay Benziger b,∗
a
College of Material Science and Engineering, State Key Laboratory for Modification of Chemical Fibers and Polymer Materials, Donghua University, Shanghai 201620, PR China
b
Chemical and Biological Engineering, Princeton University, Princeton, NJ 08544 USA

a r t i c l e i n f o a b s t r a c t

Article history: The flux of liquid water through Nafion membranes of different thickness and equivalent weight was mea-
Received 3 October 2011 sured as a function of hydrostatic pressure and temperature. Hydraulic water transport across Nafion
Received in revised form 4 December 2011 membranes increases with temperature and equivalent weight of the Nafion. Hydraulic permeability
Accepted 5 December 2011
increases with temperature due to both decreased water viscosity and increased hydrophilic volume
Available online 13 December 2011
fraction. Convective flow from the applied hydrostatic water pressure is an order of magnitude greater
than the estimated diffusive water flux associated with the water activity gradient. Water sorption and
Keywords:
hydraulic permeability data predict a hydrophilic pore network with hydrophilic domains 2.5 nm in diam-
Hydraulic permeation
Nafion
eter spaced 5.5 nm apart. The pore network structure from water sorption and hydraulic permeability
Pore network are consistent with the spacing between hydrophilic domains observed with small angle X-ray scattering
experiments.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction by a hydrophobic matrix of tetrafluoroethylene and perfluoro

Water transport through Nafion membranes is essential for the
successful operation of Polymer Electrolyte Membrane (PEM) Fuel
Cells. Water made at the cathode catalyst layer, situated between
the PEM and the porous cathode gas diffusion layer (GDL) must
be removed. Depending on the resistances to transport, water can
either flow through the porous GDL to the cathode gas flow channel
or flow through the PEM to the anode catalyst layer, anode GDL and
anode gas flow channel. Water is transported through the PEM both
by diffusion, due to an activity gradient, and by convection, due to
a hydraulic pressure difference between the cathode and anode.
In a typical PEM fuel cell there might be a pressure difference of
0.1 MPa between the cathode and the anode. There has been many
studies of water transport by diffusion through PEMs over the past
several years [1–16]. In contrast there has been only a few reports of
convective water transport through PEMs [2,5,17]. We report here
results for water uptake and convective water transport through
the PEM Nafion.
Nafion is the most common and popular PEM employed in fuel
cells. It is a perfluorosulfonic acid made by copolymerization of
tetrafluoroethylene and a perfluorovinyl ether terminated with a
sulfonyl fluoride [18]. After polymerization the sulfonyl fluoride
is hydrolyzed and ion exchanged to produce the perfluorosulfonic
acid. Nafion microphase separates into hydrophilic domains com-
prised of sulfonic acid groups and sorbed water that is surrounded
∗ Corresponding author. Tel.: +1 609 258 5416.
E-mail address: benziger@princeton.edu (J. Benziger).
ethers. Water sorption swells the hydrophilic domains providing
paths for proton transport and water diffusion [19,20]. We expect
that convective water transport driven by a hydraulic pressure will
also be through the hydrophilic domains. Several versions of Nafion
membranes are available that differ by membrane thickness, den-
sity of sulfonic acid groups and the capping groups on the ends
of the polymer chains. We anticipate that the hydraulic perme-
ation of water will correlate with the hydrophilic volume fraction
and inversely with the membrane thickness. Recent studies charac-
terizing the diffusive transport of water across Nafion membranes
with different equivalent weights (EW = mass of polymer/mole of
SO3 − X+ ) and different membrane thickness have shown that inter-
facial transport at the vapor/membrane interface can limit water
transport across thin Nafion membranes, but there is negligible
transport resistance at the liquid/membrane interface [1,2,4,20,21].
PEM fuel cells are often operated with pressure differences
between the anode and cathode; water can be transported across
the membrane by diffusion driven by a water activity difference and
by convection driven by a hydraulic pressure difference. Two recent
papers have reported hydraulic permeability of water through
Nafion was small and increased with temperature [2,4]. Kientiz
et al. examined Nafion 212 and 211 and suggested that hydraulic
permeability was small relative to diffusive transport of water [2].
We have extended the previous studies to compare water perme-
ation through Nafion membranes with different thickness, different
equivalent weight and different capping groups. The results show
that hydraulic permeability increases with temperature and vol-
ume fraction of water in the membrane. The hydraulic radius
of the water channels and their spacing were determined from

0376-7388/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2011.12.004
 Q. Duan et al. / Journal of Membrane Science 392–393 (2012) 88–94
Fig. 1. Schematic of flow cell for water convection through Nafion membranes.
measurements of water volume fraction and water flow rate in a
porous network. The distance between hydrophilic domains was
∼5.5 nm in agreement with the characteristic length observed by
scattering experiments [18].
2. Experimental
2.1. Flow experiments
Liquid water flow rates were measured with the permeation
cell shown schematically in Fig. 1. Two polycarbonate blocks
were machined with plenums 2.5 cm diameter × 1.25 cm deep. The
blocks were bolted together with a polymer membrane placed
between the two plenums. A support cross was fitted in the low
pressure plenum to limit deflection of the polymer membrane.
Water was fed to one plenum from a pressurized reservoir. The
hydrostatic pressure from the height of the water reservoir above
the membrane and the applied gas pressure were combined to
give a total hydraulic pressure. The outlet from the second plenum
went into a 1.6 mm ID tygon tube. The water flow rate across the
membrane was determined from the distance the water moved
through the tygon tube as a function of time. Water flow rates
were measured at a fixed hydraulic pressure for a period of 2–3 h to
obtain steady state flow rates. Flow rates were measured for both
increasing and decreasing pressure changes. The permeation cell
was placed inside an insulated heated box where the temperature
was controlled.
2.2. Membrane preparation and treatment
All membranes were treated by boiling in 3% H2 O2 for 1 h, rins-
ing in boiling deionized water for 20 min, boiling in 1 M H2 SO4
for an hour and finally rinsing again in boiling deionized water
for 30 min. Commercial extruded Nafion 115, Nafion 1110, Nafion
1035 and Nafion 212 were obtained from Ion Power (New Castle
DE) and treated as outlined above. Nafion 115 and Nafion 1110 are
both equivalent weight 1100 with dry thicknesses 127 and 254 ␮m,
respectively. Nafion 212 is equivalent weight 1100 with a dry thick-
ness of 51 ␮m. Nafion 212 has a different capping group on the
polymer chain than Nafion 115 and Nafion 1110. Nafion 1035 is
equivalent weight 1000 with a dry membrane thickness of 89 ␮m.
2.3. SAXS sample preparation
Small angle X-ray scattering was carried out on samples of
Nafion 1110 equilibrated with water vapor and liquid water at
room temperature and a sample placed in boiling water. The
89
Fig. 2. Water flux through Nafion 115 membranes at different temperatures.
membranes were placed in a sealed sample cell constructed from
two copper plates with mica windows and a Viton gasket between
them. Membrane water contents, , were determined gravimet-
rically on samples having the same treatment as those being
mounted in the SAXS cell.
The sample cell was placed in the path of the X-ray beam. In
order to minimize membrane water loss during the data collec-
tion, helium gas was used in the flight tube rather than a vacuum.
The 1.54 Å Cu-K␣ X-rays were generated by a Philips XRG-3000
sealed tube generator source. The beam was slit collimated and the
scattering was detected by an Anton-Paar compact Kratky cam-
era equipped with a Braun OED-50M detector. Samples were run
at room temperature for 10 min. Empty beam scattering, sample
transmittance, and detector response were corrected for in the
data analysis. The data reduction and desmearing procedures are
described in detail by Register [22]. The invariant scattering inten-
sity (q2 I) is plotted as a function of scattering angle or distance
and the Bragg spacing is determined by the location of the peak in
scattering intensity.
2.4. Water uptake
The equilibrium water uptake was measured as a function of
water pressure at different temperatures using an isometric system
[19,23]. A PEM is placed in a fixed volume container. The PEM sam-
ple was evacuated to below 1 Pa (<0.01 mbar) at 80 ◦ C to remove
all the water from the membrane. The pressure container was then
cooled to the desired temperature under vacuum. The valve to the
vacuum line was shut off and aliquots of water are injected into
the pressure vessel and allowed to equilibrate with the Nafion. The
resulting pressure is equal to the partial pressure of water. The dif-
ference between the quantity of water injected and the water in the
gas phase is equal to the amount of water absorbed by the Nafion.
Saturation was evident when the injection of a liquid aliquot did
not cause any change in the pressure.
3. Results
The water fluxes as functions of applied hydraulic pressure for
Nafion 115 at room temperature (∼23 ◦ C), 40, 60 and 80 ◦ C are
shown in Fig. 2. Data for different Nafion membranes, N115, N1110,
N1035 and N212 at room temperature are shown in Fig. 3. These
data are all well represented by a linear increase of water flux with
applied hydraulic pressure. Table 1 summarizes the slope of the
90 Q. Duan et al. / Journal of Membrane Science 392–393 (2012) 88–94

Table 1
Nafion membrane hydraulic permeabilities and water sorption.

Membrane (EW) T (◦ C)  (Pa s) tm (␮m) kw (cm2 × 1016 ) sat = #H2 O/SO3

N115 (1100) 23 .100 127 3.75 21 ± 2

N115 (1100) 40 .065 127 5.04 22 ± 2
N115 (1100) 60 .047 127 5.49 24 ± 2
N115 (1100) 80 .035 127 7.13 25 ± 2
N1110 (1100) 23 .100 254 3.90 21 ± 2
N1035 (1000) 23 .100 89 5.77 21 ± 2
N212 (1100) 23 .100 51 4.09 21 ± 2

water flux (Qw ) vs. pressure (P) for the different samples and dif-
ferent temperatures. Membrane permeabilities (kw ), defined by Eq.
(1), were calculated from the flux data and are given in Table 1. The
membrane permeability is expected to be independent of water
viscosity, , and membrane thickness, tm .
increase in water sorption between saturated vapor and liquid, and
an additional small increase between saturated liquid at 20 ◦ C and
100 ◦ C. The characteristic scattering length increases slightly with
increased water content in the membrane.

kw
Qw = P
tm
The results indicate that the three membranes with the 1100
equivalent weight, N115, N1110 and N212, all had similar perme-
abilities circa 4 × 10−16 cm2 at 23 ◦ C. N1035, with an equivalent
weight of 1000, had permeability about 40% greater than the
1100 EW Nafion. The permeability of N115 membranes increased
with temperature. Since kw corrects for the viscosity change with
temperature, the data show that the permeability increased with
temperature.
The saturation uptake of water into the Nafion membranes
(reported as  = number of water molecules sorbed per sulfonate
group) are shown in Table 1. The saturation water volume increased
modestly with increasing temperature. The saturation water vol-
ume at 23 ◦ C was also larger for the 1000 EW Nafion (N1035) than
the 1100 EW Nafion. (Even though  was the same for N115 and
N1035 the water volume fraction is larger for N1035 because the
density of sulfonic acid groups is greater for N1035.)
Small angle X-ray scattering results for four different Nafion
1110 samples are shown in Fig. 4. More detailed studies can be
found in a number of sources [18,24–27]. The purpose of Fig. 4 is
to highlight the characteristic scattering length of ∼5 nm for Nafion
samples with absorbed water. There is no evidence of any scattering
in the nm range for proton exchanged Nafion membranes with-
out absorbed water. The other three samples correspond to water
absorption at saturation conditions: saturated vapor at room tem-
perature, saturated liquid at room temperature and saturated liquid
at 100 ◦ C. The SAXS results for 1100 EW Nafion indicate a small
4. Discussion
(1)
The mechanisms of water transport in Nafion and structure
of the “water channels” have been a source of intense interest
over the past 25 years. Modeling efforts have examined different
contributions of water sorption and transport including diffusion
and convection. Eikerling and co-workers have considered various
driving forces for convection including external pressure gradient,
capillary pressure, osmotic pressure, and elastic forces associated
with membrane deformation [28–31]. Experiments have lagged
behind the modeling efforts for water transport and only recently
have experimental results been published that provide details
of transport resistances [1,4,5,10,20,32,33], mechanical properties
[34,35] and water profiles [36–39] as functions of water activity and
temperature. These recent experimental results can be compared
with previous modeling efforts to improve the understanding of
structure–property relationships in Nafion.
Water permeation, water sorption and SAXS studies of water in
Nafion have been reported in the literature [18]. It is well estab-
lished that water sorption in Nafion is accompanied by a strong
scattering peak with a characteristic distance of ∼5 nm. Water dif-
fusion increases with water content in Nafion. However, there
has been minimal quantitative connections between the struc-
ture of Nafion determined by scattering experiments and transport
rates. We explore here the consistency of convective water trans-
port through Nafion and the structure of the hydrophilic domains
determined from scattering experiments. It will be shown that con-
vective water transport coupled with water sorption predicts water

Fig. 4. Small angle X-ray scattering of Nafion 1110 membranes in saturated vapor,
Fig. 3. Water flux through different Nafion Membranes at 23 ◦ C. room temperature saturated liquid and saturated liquid at 100 ◦ C.
 Q. Duan et al. / Journal of Membrane Science 392–393 (2012) 88–94 91

is convected through a hydrophilic “channel” network consisting of a substantial hydrostatic pressure difference (∼100 kPa) between
2.5 nm diameter water channels spaced 5.5 nm apart; this structure the anode and cathode.
shows quantitative agreement with the results from SAXS. Benziger and co-workers, and Monroe et al. concluded that
interfacial water transport at the membrane vapor interface is
4.1. Effect of temperature and equivalent weight on hydraulic the principal resistance to water transport in pervaporation sys-
permeability tems [1,8,20]. Because the limiting transport resistance is at the
vapor/membrane interface, the water activity in the membrane is
Two recent papers have reported measurements of pressure nearly constant across the membrane, approximately equal to the
driven water convection through Nafion. Weber and collaborators liquid water activity, as shown by Eq. (3), where km is a function of
from Toyota reported permeability coefficients for Nafion 211 and temperature.
Nafion 212 [2]. Their values for 1100 EW Nafion is in the range vapor liquid vapor
(Qpervaporation ) = km (amembrane − aw ) ≈ km (aw − aw ) (3)
of 1–8 × 10−16 cm2 very close to those reported here at 60 and w

80 ◦ C. The permeability they measured at 25 ◦ C was 4 times smaller Saturated liquid activity is unity (aw
saturated liquid
= 1). The activity
than the value we report. Holdcroft and co-workers also measured correction for a pressurized liquid is given by the Poynting correc-
hydraulic water permeability through Nafion 212 [4,5,21]. They tion shown in Eq. (4) (assuming that liquid water is incompressible
also obtained values circa 10−16 cm2 similar to those reported here. over the pressures of interest).
These recent values are almost 100 times less than earlier values    
reported by Bernaradi and Verbrugge [17]. We suspect there were P sat )V̄
sat V̄w (P − Pw w
some difficulties with those early experiments. aw = aw (Pw ) + exp dP = 1 + exp
sat RT RT
Hydraulic permeabilities at 23 ◦ C were almost the same for Pw

N115, N1110 and N212, all of which have an equivalent weight
of 1100. These three membranes also sorbed the same number of
water molecules per sulfonic acid groups ( = 21H2 O/SO3 at 23 ◦ C).
The hydraulic permeability of N1035, with EW 1000, is greater than
the permeability of the 1100 EW Nafion membranes.
Our results, as well as those of Kientiz et al. and Adachi et al.,
found that the hydraulic permeability increased with tempera-
ture. In all three studies the water flux increased more than that
predicted by the decrease of water viscosity with temperature, indi-
cating that the hydraulic permeability increased with temperature.
Results in Table 1 also show that the number of water molecules
sorbed per sulfonic acid group increased with temperature, indi-
cating that the volume fraction of water in Nafion increased with
increasing temperature. The increase in hydraulic permeability
with temperature and equivalent weight is greater than expected
from the increased volume fraction of water. The water volume
fraction, ϕw , given in Eq. (2) changes by almost 10% while the
hydraulic permeability increased by 50–80%. This difference sug-
gests there may be structural changes to the hydrophilic domains
with increased water sorption beyond simple volume expansion.
(Nafion /EW)Vw
ϕw =
1 + (Nafion /EW)Vw
4.2. Hydraulic permeation vs. pervaporation
The recent paper by Kientiz et al. compared the water fluxes
from liquid/vapor pervaporation and liquid/liquid hydraulic per-
meation. They observed that the liquid/vapor pervaporation fluxes
were significantly greater than the liquid/liquid hydraulic perme-
ation fluxes. Kientiz et al. suggested that hydraulic permeation
could be neglected relative to the pervaporation if there was an
activity gradient [2]. Adachi et al. also made similar observations
[4]. In our liquid/vapor permeation experiments in our labs we
also observed larger pervaporation water fluxes than the convec-
tive water fluxes we report here [1,20,32]. Is hydraulic permeation
important to water transport in Nafion?
Water transport across Nafion membranes is essential to both
membrane humidifiers and PEM fuel cells. The water activity dif-
ferences across the membrane are quite different in fuel cells
compared to membrane humidifiers. Nafion membrane humid-
ifiers operate as pervaporation cells with a small hydrostatic
pressure difference (<10 kPa) but large relative humidity difference
across the membrane [40–42]. In contrast, fuel cells have saturated
liquid/gas streams at both sides of the membrane throughout most
of the fuel cell; there is a negligible relative humidity difference, but
sat )V̄
(P − Pw w
≈1+ (4)
RT
The activity correction to pressurized water at 100 ◦ C from satu-
ration (0.1 MPa) to 1 MPa pressure (10 bar) is ∼0.001. Hence, we
conclude that at low to moderate differential pressures no mea-
sureable change to the water pervaporation flux is expected. This
is consistent with pressurized pervaporation experiments by Kien-
tiz et al. and Adachi et al. (We did similar experiments. Initially
we observed a small change in the steady state flux for pressur-
ized pervaporation through Nafion. However, when we followed
the flux dynamically we discovered the change in flux was due to
bowing of the membrane due to the pressure differential, which
increased the interfacial area.)
There is no convective water flux across the membrane/vapor
interface. Since interfacial transport at the vapor/membrane inter-
face is the limiting resistance to pervaporation, any contribution
to water transport by hydraulic permeation in the membrane will
increase the water concentration at the membrane/vapor inter-
face and reduce the already small water activity gradient across
the membrane.
When liquid water is at both membrane surfaces the interfa-
(2)
cial transport resistances are negligible. Unless there is a pressure
(or temperature) gradient across the membrane there is no driv-
ing force for water transport. A differential pressure across the
membrane will drive water transport by diffusion and hydraulic
permeation. Diffusive permeation is driven by the water activity
gradient across the membrane. The activity difference for pressur-
ized liquid water on both sides of the membrane is given by the
Poynting correction shown in Eq. (5).
 Pcathode     
V̄w (Panode − Pcathode )V̄w
aw = exp dP = exp
Panode
RT RT
(Panode − Pcathode )V̄w
≈ (5)
RT
The diffusive water flux (Qw,diffusive ) associated with the hydro-
static pressure difference across the membrane of thickness tm , is
given by Eq. (6); where Dw is the diffusion coefficient of water and
cw is the concentration of water in the membrane (cw = NSO3 =
Nafion /EW where NSO3 is the density of sulfonic acid groups in
Nafion). Diffusion coefficients for 1100 EW Nafion were reported
by Zhao et al. [20]; Dw (aw = 1) ≈ 5 × 10−6 cm2 s−1 .
aw Dw NSO3 (Panode − Pcathode )V̄w
Qw,diffusive ≈ Dw cw,mem = (6)
tm tm RT
92 Q. Duan et al. / Journal of Membrane Science 392–393 (2012) 88–94
The ratio of the diffusive flux (Eq. (6)) to the convective flux (Eq.
(1)) with liquid present on both sides of the membrane is given by
Eq. (7).
Qdiffusive Dw NSO3 V̄w
= ≈ 3 × 10−2
Qconvective RTkw
When water is present at both sides of the Nafion membrane the
diffusive flux from an applied pressure is only a small fraction of the
hydraulic flux (∼3%). In PEM fuel cells when the anode and cathode
gas streams are both fully humidified the transport of water across
Nafion membranes will be dominated by hydraulic permeation.
4.3. Water removal from the catalyst layer: through the
membrane or through the GDL?
Water is produced at the cathode catalyst layer. The vapor is
saturated in the cathode flow channel. There is no water activ-
ity gradient to drive diffusion from the cathode catalyst layer to
the cathode gas flow channel, so liquid water will accumulate in
the catalyst layer and build up a hydrostatic pressure. The water
pressure will increase until the water pushes through either the
membrane or the cathode gas diffusion layer. The pressure required
to push water through the Nafion membrane depends on the cur-
rent density of the fuel cell. The hydrostatic pressure to remove
the water by transport through the membrane is found by equat-
ing the water flux through the membrane to the water production.
The hydrostatic pressure differential to force the water through the
membrane at typical fuel cell current densities would be very large
(equation 8).
j/2F
P = ≈ 1 MPa, j = 1 A/cm2
kw /tm
The hydraulic pressure to force water through the membrane can
be compared to the pressure to force liquid water through the GDL.
Benziger and co-workers previously measured liquid water per-
meation through GDL materials [43]. There is a minimum pressure
for liquid water penetration into the largest pores of the GDL of
∼10 kPa. Above the penetration pressure the permeation coeffi-
cient through the GDL is ∼10−12 cm2 , which is much greater than
kw ∼ 10−16 cm2 for permeation through the membrane. This simple
calculation shows that most of the water formed at the cathode will
be transported through the GDL to the cathode gas flow channel.
4.4. The hydrophilic pore network
Nafion consists of hydrophilic sulfonic acid groups dispersed
in a hydrophobic matrix of tetrafluoroethylene and perfluoroalkyl
ethers [18,24,44–53]. The sulfonic acid groups sorb water swelling
the hydrophilic network through which water and protons are
transported. The microphase separated structure of Nafion is analo-
gous to a porous medium where water may be transported through
a convoluted hydrophilic pore structure. Gierke proposed a pore
structure with spherical hydrophilic clusters of sulfonic acid groups
and water that are connected by narrow hydrophilic channels
[44]. More refinements to the Gierke model have been proposed
over the past 30 years based primarily on results from Small
Angle X-ray scattering and Small Angle Neutron scattering [18].
Those results have shown that there is a characteristic scattering
distance of ∼5 nm that appears when water is sorbed in Nafion
[24,27,48–51]. Most studies have associated that distance with the
spacing between hydrophilic clusters in the Nafion [24,48,49,54];
however, some studies have suggested that the scattering peak
corresponds to the size of the hydrophilic clusters [55,56].
Transport properties can also provide structural information
about Nafion. Wu et al. have recently shown that there is a crit-
ical hydrophilic volume fraction (ϕ) threshold of 0.1 in 1100
(7)
Fig. 5. Cell model for hydrophilic network structure. (A) This is the structure of
Nafion predicted by molecular dynamics with  = 15H2 O/SO3 . Only the water and
sulfonic acid groups are shown. The voids are occupied by the TFE backbone and
PFEA side chains. (B) A unit cell of a simplified model for Nafion assuming a geo-
metrically regular structure. The hydrophilic domains are the struts and the voids
are the hydrophobic domains. The hydraulic radius for water transport is given by a,
and d is the spacing between hydrophilic domains. (C) The connection of unit cells
to make a bicontinuous channel network.
EW Nafion for the hydrophilic domains to percolate through the
hydrophobic matrix [19]. Above the critical threshold, proton
conductivity increases quadratically with the hydrophilic volume
fraction; H+ ∼(ϕ − ϕc )2 . Zhao et al. have shown that water diffu-
sion in Nafion also increases quadratically with the hydrophilic
(8)
volume fraction; Dw ∼ ϕ2 [20]. Zhao also showed that tortuosity,
, of the hydrophilic domains decreases dramatically with water
volume fraction from  ∼ 20 at ϕw = 0.02 to  ∼ 2 at ϕw ∼ 0.1.
The convection of water through Nafion can also serve as a probe
to its structure. The hydrophilic domains form a network through
which water flows. The water flux depends on the size and den-
sity of the hydrophilic channels. Fig. 5A shows molecular dynamics
results from Daly for the hydrophilic phase separated domains for
1100 EW Nafion with  = 15H2 O/SO3 [57]. The hydrophilic domains
form a channel network. A simplified model of the channel network
in Nafion is illustrated in Fig. 5. This model is greatly simplified,
assuming regularly spaced uniform channels. The utility of the
model is to make quantitative predictions of flow rates through
Nafion.
The microphase separated structure is approximated by a reg-
ular array of unit cells consisting of struts and cavities. The
struts and cavities of the unit cell represent Nafion’s phase sep-
arated structure. The volume fractions of the struts and cavities
are characterized by two quantities: a = characteristic dimen-
sion of the hydrophilic channel (e.g. hydrodynamic radius), and
d = characteristic distance between hydrophilic channels. The vol-
ume fraction of the struts, representing water sorbed into the
hydrophilic channels is related to the two quantities a and d by Eq.
(9). (One could imagine the hydrophilic and hydrophobic phases
switched. Because the model is symmetric the results are indepen-
dent of whether one assumed the hydrophilic phases consisted of
the struts between vertices or the central voids.)
 a 2  a 3
ϕ = 12 − 16 (9)
d d
The quadratic term in Eq. (9) is the volume of non-intersecting
channels; the cubic term in Eq. (9) accounts for the overlapping
volumes of channels going in different directions.
When a hydraulic pressure is applied it will cause water to flow
in the direction of the pressure gradient; the applied pressure must
 Q. Duan et al. / Journal of Membrane Science 392–393 (2012) 88–94 93

Table 2
Pore network parameters for liquid water flow in Nafion.

Membrane T (◦ C) ϕw [2kw ]1/2 (nm) a (nm) d (nm)

N115 23 0.404 ± 0.03 0.27 1.12 ± 0.15 5.14 ± 0.2

N115 40 0.36 ± 0.03 0.32 1.26 ± 0.15 5.69 ± 0.2
N115 60 0.37 ± 0.03 0.33 1.24 ± 0.15 5.43 ± 0.2
N115 80 0.39 ± 0.03 0.38 1.38 ± 0.15 5.94 ± 0.2
N1110 23 0.34 ± 0.03 0.27 1.14 ± 0.15 5.25 ± 0.2
N1035 23 0.38 ± 0.03 0.34 1.28 ± 0.15 5.57 ± 0.2
N212 23 0.35 ± 0.03 0.28 1.17 ± 0.15 5.37 ± 0.2

overcome the viscous drag with the walls of the channels. The
flow rate as a function of pressure drop may be approximated by
Poiseuille flow (Eq. (10)), where laminar flow in uniform square
channels with hydraulic radius rh is assumed.
 
rh2 P
Qw,pore = (2rh )2
8 Lpore
The total water flow for channels spaced at a distance d apart is
given by Eq. (11). To account for distortions of the real pore network
from the idealized model the effective length of the channels is
given by the membrane thickness multiplied by a tortuosity factor,
.
   
The results indicate that the spacing between hydrophilic
domains, d, ranges from 5 to 6 nm, increasing with water
volume fraction. Those values agree with the characteristic dis-
tance observed from scattering experiments (e.g. see Fig. 4).
The hydrophilic channel diameter, dchannel = 2a, is approximately
(10) 2.3–2.7 nm.
The pore structure suggested here is similar to that previous
proposed by Gebel and Diat for proton transport in Nafion [48,51].
Gebel and Lambard had previously combined water sorption data
with the characteristic dimension from SAXS experiments to pre-
dict hydrophilic domain sizes and spacing [51]. The agreement of
the water sorption data and Porod analysis from Gebel and co-

workers was not very good assuming spherical domains. The fibril
(2rh )2 rh2 P rh4 P structure of hydrophobic cores surrounded by a hydrophobic shell
Qw = = (11)
d2 8 Lpore d2 2tm more recently proposed by Gebel and co-workers was more con-
sistent with scattering results and sorption data and we show here
Eqs. (10) and (11) assume that the Poiseuille flow description of it is also consistent with transport data.
liquid flow through cylindrical channels can be extended to more The water sorption and hydraulic water transport are not con-
complex geometries, this assumption has been validated for both sistent with models that suggest the hydrophilic pores are ∼5 nm
gas and liquid flow through porous materials, leading to the com- in diameter [56]. The flow results are also not consistent with the
monly employed Kozeny equation for flow through porous media classical Gierke model of spherical clusters connected by narrow
[58]. channels. If there were narrow channels between large spher-
The hydraulic radius accounts for the volume available for flow ical clusters the hydraulic permeability of the pores would be
divided by interfacial area over which the shear forces are exerted, greatly reduced, which would require the spacing between pores
as given by Eq. (12). The hydrophilic volume is the volume fraction to be reduced. The flow rate at fixed pressure increases with the
of water sorbed into the polymer membrane, given by Eq. (2). hydraulic radius raised to the 4th power, so a small decrease in
hydrophilic volume/unit volume ϕw channel radius would require more closely spaced hydrophilic
rh = = (12) channels.
interfacial area/unit volume ˛int
The interfacial area between the hydrophilic domains and the
hydrophobic matrix is not directly accessible for measurement.
However, the interfacial area is related to the hydrophilic network
parameters a and d, and is given by Eq. (13). Eq. (13) only includes 5. Conclusions
the interfacial area of the struts in the direction of flow.
Hydraulic permeation of water through Nafion membranes has
˛int = 24a(d − 2a) (13) been examined for different equivalent weights at temperatures
20–80 ◦ C. The principal findings are:
The hydraulic radius can also be expressed as a function of the
hydrophilic network parameters as given by Eq. (14).

3ad − 4a2
rh = (14) 1. Hydraulic permeation decreases with increasing membrane
3(d − 2a)
thickness.
Eqs. (2) and (11)–(14) can be combined to determine a and d 2. Hydraulic permeation increases with increasing temperature.
as functions of the experimentally measured and water volume 3. Hydraulic permeation increases with decreasing equivalent
fraction (ϕw ) and water flow rate (Qw ). weight of Nafion.
The water volume fractions were determined from water sorp- 4. Hydraulic permeation is of negligible importance for water per-
tion data and the results are listed in Table 2. From Eqs. (1) and vaporation; the increase of liquid water activity with pressure is
(11) the experimental permeability can be related to the hydraulic small.
radius and spacing between pores. 5. Hydraulic permeation is the dominant transport process for
water when liquid water is present on both sides of a Nafion
rh4 membrane.
= [2kw ] (15)
d2 6. Liquid water removal from the cathode catalyst layer is primarily
Assuming  = 2 from the measurements of Zhao et al. [20] values through the cathode GDL.
of a and d were determined for each membrane at the different 7. Water sorption and hydraulic permeation results suggest a
temperatures using Eqs. (9) and (15); the results are summarized hydrophilic pore network in Nafion with hydrophilic domains
in Table 2. ∼2.5 nm in diameter spaced 5.5 nm apart.
94 Q. Duan et al. / Journal of Membrane Science 392–393 (2012) 88–94
Acknowledgements
This work was supported by the National Science Foundation
MRSEC Program through the Princeton Center for Complex Materi-
als NSF DMR-0819860. Q.D. thanks the Chinese Scholarship Council
for fellowship support.
References
[1] P.W. Majsztrik, M.B. Satterfield, A.B. Bocarsly, J.B. Benziger, Water sorption,
desorption and transport in Nafion membranes, Journal of Membrane Science
301 (2007) 93–106.
[2] B. Kientiz, H. Yamada, N. Nonoyama, A.Z. Weber, Interfacial water transport
effects in proton-exchange membranes, Journal of Fuel Cell Science and Tech-
nology 8 (2011).
[3] S.H. Ge, X.G. Li, B.L. Yi, I.M. Hsing, Absorption, desorption, and transport of water
in polymer electrolyte membranes for fuel cells, Journal of The Electrochemical
Society 152 (2005) A1149–A1157.
[4] M. Adachi, T. Navessin, Z. Xie, B. Frisken, S. Holdcroft, Correlation of in situ
and ex situ measurements of water permeation through Nafion NRE211 pro-
ton exchange membranes, Journal of the Electrochemical Society 156 (2009)
B782–B790.
[5] M. Adachi, T. Navessin, Z. Xie, F.H. Li, S. Tanaka, S. Holdcroft, Thickness depen-
dence of water permeation through proton exchange membranes, Journal of
Membrane Science 364 (2010) 183–193.
[6] V. Freger, E. Korin, J. Wisniak, E. Korngold, M. Ise, K.D. Kreuer, Diffusion of water
and ethanol in ion-exchange membranes: limits of the geometric approach,
Journal of Membrane Science 160 (1999) 213–224.
[7] J.T. Hinatsu, M. Mizuhata, H. Takenaka, Water-uptake of perfluorosulfonic acid
membranes from liquid water and water-vapor, Journal of The Electrochemical
Society 141 (1994) 1493–1498.
[8] C.W. Monroe, T. Romero, W. Merida, M. Eikerling, A vaporization-exchange
model for water sorption and flux in Nafion, Journal of Membrane Science 324
(2008) 1–6.
[9] S. Motupally, A.J. Becker, J.W. Weidner, Water transport in polymer electrolyte
membrane electrolyzers used to recycle anhydrous HCl. I. Characterization of
diffusion and electro-osmotic drag, Journal of The Electrochemical Society 149
(2002) D63–D71.
[10] M.B. Satterfield, J.B. Benziger, Non-fickian water vapor sorption dynamics by
Nafion membranes, Journal of Physical Chemistry B 112 (2008) 3693–3704.
[11] T. Takamatsu, M. Hashiyama, A. Eisenberg, Sorption phenomena in Nafion
membranes, Journal of Applied Polymer Science 24 (1979) 2199–2220.
[12] S. Tsushima, K. Teranishi, S. Hirai, Water diffusion measurement in fuel-cell SPE
membrane by NMR, Energy 30 (2005) 235–245.
[13] A.Z. Weber, J. Newman, Transport in polymer-electrolyte membranes. I. Phys-
ical model, Journal of The Electrochemical Society 150 (2003) A1008–A1015.
[14] Q.G. Yan, H. Toghiani, J.X. Wu, Investigation of water transport through mem-
brane in a PEM fuel cell by water balance experiments, Journal of Power Sources
158 (2006) 316–325.
[15] T.A. Zawodzinski, C. Derouin, S. Radzinski, R.J. Sherman, V.T. Smith, T.E.
Springer, S. Gottesfeld, Water-uptake by and transport through Nafion(R) 117
membranes, Journal of The Electrochemical Society 140 (1993) 1041–1047.
[16] T.A. Zawodzinski, M. Neeman, L.O. Sillerud, S. Gottesfeld, Determination of
water diffusion-coefficients in perfluorosulfonate ionomeric membranes, Jour-
nal of Physical Chemistry 95 (1991) 6040–6044.
[17] M.W. Verbrugge, E.W. Schneider, R.S. Conell, R.F. Hill, The effect of
temperature on the equilibrium and transport-properties of saturated
poly(perfluorosulfonic acid) membranes, Journal of The Electrochemical Soci-
ety 139 (1992) 3421–3428.
[18] K.A. Mauritz, R.B. Moore, State of understanding of Nafion, Chemical Reviews
104 (2004) 4535–4585.
[19] X. Wu, X. Wang, G. He, J. Benziger, Differences in water sorption and proton
conductivity between Nafion and SPEEK, Journal of Polymer Science Part B:
Polymer Physics 49 (2011) 1437–1445.
[20] Q. Zhao, P. Majsztrik, J. Benziger, Diffusion and interfacial transport of water in
Nafion, Journal of Physical Chemistry B 115 (2011) 2717–2727.
[21] J. Peron, A. Mani, X.S. Zhao, D. Edwards, M. Adachi, T. Soboleva, Z.Q. Shi, Z. Xie, T.
Navessin, S. Holdcroft, Properties of Nafion (R) NR-211 membranes for PEMFCs,
Journal of Membrane Science 356 (2010) 44–51.
[22] R.A. Register, T.R. Bell, Miscible blends of zinc-neutralized sulfonated
polystyrene and poly(2,6-dimethyl 1,4-phenylene oxide), Journal of Polymer
Science Part B: Polymer Physics 30 (1992) 569–575.
[23] C. Yang, S. Srinivasan, A.B. Bocarsly, S. Tulyani, J.B. Benziger, A comparison of
physical properties and fuel cell performance of Nafion and zirconium phos-
phate/Nafion composite membranes, Journal of Membrane Science 237 (2004)
145–161.
[24] G. Gebel, O. Diat, Neutron and X-ray scattering: suitable tools for studying
ionomer membranes, Fuel Cells 5 (2005) 261–276.
[25] K.A. Page, F.A. Landis, A.K. Phillips, R.B. Moore, SAXS analysis of the thermal
relaxation of anisotropic morphologies in oriented Nafion membranes, Macro-
molecules 39 (2006) 3939–3946.
[26] L. Rubatat, O. Diat, Stretching effect on Nafion fibrillar nanostructure, Macro-
molecules 40 (2007) 9455–9462.
[27] M. Pineri, G. Gebel, R.J. Davies, O. Diat, Water sorption–desorption in Nafion (R)
membranes at low temperature, probed by micro X-ray diffraction, Journal of
Power Sources 172 (2007) 587–596.
[28] M. Eikerling, Y.I. Kharkats, A.A. Kornyshev, Y.M. Volfkovich, Phenomenological
theory of electro-osmotic effect and water management in polymer electrolyte
proton-conducting membranes, Journal of the Electrochemical Society 145
(1998) 2684–2699.
[29] J. Divisek, M. Eikerling, V. Mazin, H. Schmitz, U. Stimming, Y.M. Volfkovich, A
study of capillary porous structure and sorption properties of Nafion proton-
exchange membranes swollen in water, Journal of The Electrochemical Society
145 (1998) 2677–2683.
[30] M. Eikerling, A.A. Kornyshev, U. Stimming, Electrophysical properties of poly-
mer electrolyte membranes: a random network model, Journal of Physical
Chemistry B 101 (1997) 10807–10820.
[31] M.H. Eikerling, P. Berg, Poroelectroelastic theory of water sorption and swelling
in polymer electrolyte membranes, Soft Matter 7 (2011) 5976–5990.
[32] P. Majsztrik, A. Bocarsly, J. Benziger, Water permeation through nafion mem-
branes: the role of water activity, Journal of Physical Chemistry B 112 (2008)
16280–16289.
[33] M. Adachi, T. Navessin, Z. Xie, S. Holdcroft, POLY 364-Ex-situ water permeation
measurements through proton exchange membranes, Abstracts of Papers of
the American Chemical Society 238 (2009).
[34] P.W. Majsztrik, A.B. Bocarsly, J.B. Benziger, Viscoelastic response of nafion.
effects of temperature and hydration on tensile creep, Macromolecules 41
(2008) 9849–9862.
[35] M.B. Satterfield, J.B. Benziger, Viscoelastic properties of nafion at elevated tem-
perature and humidity, Journal of Polymer Science Part B: Polymer Physics 47
(2009) 11–24.
[36] G. Gebel, O. Diat, S. Escribano, R. Mosdale, Water profile determination in a
running PEMFC by small-angle neutron scattering, Journal of Power Sources
179 (2008) 132–139.
[37] G. Gebel, S. Lyonnard, H. Mendil-Jakani, A. Morin, The kinetics of water sorp-
tion in Nafion membranes: a small-angle neutron scattering study, Journal of
Physics-Condensed Matter 23 (2011).
[38] Z. Zhang, A.E. Marble, B. MacMillan, K. Promislow, J. Martin, H. Wang, B.J.
Balcom, Spatial and temporal mapping of water content across Nafion mem-
branes under wetting and drying conditions, Journal of Magnetic Resonance
194 (2008) 245–253.
[39] Z. Zhang, K. Promislow, J. Martin, H. Wang, B.J. Balcom, Bi-modal water trans-
port behavior across a simple Nafion membrane, Journal of Power Sources 196
(2011) 8525–8530.
[40] S.K. Park, E.A. Cho, I.H. Oh, Characteristics of membrane humidifiers for polymer
electrolyte membrane fuel cells, Korean Journal of Chemical Engineering 22
(2005) 877–881.
[41] Perma Pure LLC, http://www.permapure.com/industry/enviro/technology/,
2011.
[42] Cell Craft, http://www.cellkraft.se/humidity and steam/home.html, 2011.
[43] J. Benziger, J. Nehlsen, D. Blackwell, T. Brennan, J. Itescu, Water flow in the
gas diffusion layer of PEM fuel cells, Journal of Membrane Science 261 (2005)
98–106.
[44] T.D. Gierke, W.Y. Hsu, The cluster-network model of ion clustering in perfluo-
rosulfonated membranes, ACS Symposium Series 180 (1982) 283–307.
[45] T.D. Gierke, G.E. Munn, F.C. Wilson, The morphology in Nafion perfluori-
nated membrane products, as determined by wide-angle and small-angle
X-ray studies, Journal of Polymer Science Part B: Polymer Physics 19 (1981)
1687–1704.
[46] T.D. Gierke, G.E. Munn, F.C. Wilson, Morphology of perfluorosulfonated mem-
brane products – wide-angle and small-angle X-ray studies, ACS Symposium
Series 180 (1982) 195–216.
[47] A. Eisenberg, Clustering of ions in organic polymers. A theoretical approach,
Macromolecules 3 (1970) 147–154.
[48] O. Diat, G. Gebel, Proton channels, Nature Materials 7 (2008) 13–14.
[49] G. Gebel, P. Aldebert, M. Pineri, Swelling study of perfluorosulphonated
ionomer membranes, Polymer 34 (1993) 333–339.
[50] G. Gebel, Structural evolution of water swollen perfluorosulfonated ionomers
from dry membrane to solution, Polymer 41 (2000) 5829–5838.
[51] G. Gebel, J. Lambard, Small-angle scattering study of water-swollen perfluori-
nated ionomer membranes, Macromolecules 30 (1997) 7914–7920.
[52] K.D. Kreuer, On the development of proton conducting polymer membranes
for hydrogen and methanol fuel cells, Journal of Membrane Science 185 (2001)
29–39.
[53] M. Pineri, A. Eisenberg, Structure and properties of ionomers, in: NATO ASI
Series. Series C: Mathematical and Physical Sciences, Reidel Pub. Co., Norwell,
MA, 1987.
[54] K. Schmidt-Rohr, Q. Chen, Parallel cylindrical water nanochannels in Nafion
fuel-cell membranes, Nature Materials 7 (2008) 75–83.
[55] M. Fujimura, T. Hashimoto, H. Kawai, Small-angle X-ray-scattering study of
perfluorinated ionomer membranes. 1. Origin of 2 scattering maxima, Macro-
molecules 14 (1981) 1309–1315.
[56] M. Fujimura, T. Hashimoto, H. Kawai, Small-angle X-ray-scattering study of
perfluorinated ionomer membranes. 2. Models for ionic scattering maximum,
Macromolecules 15 (1982) 136–144.
[57] K. Daly, Structure and Properties of Nafion from Molecular Simulations, Prince-
ton University, Princeton, NJ, 2011.
[58] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena, 2nd ed., Wiley,
New York, 2002.